MTA's A.I. bus cameras issue thousands of mistaken parking violations

It's one thing to get a single parking violation. It is quite another to get three in the same day.

So when George Han received a trio of violations in the mail for blocking the bus lane on East 79th Street, he scratched his head. He was certain he didn’t break New York City’s parking rules.

“At first I wasn’t worried, because I know the parking rules and these are legal spots,” he said. “It turns out that the tickets kept coming though. We are up to about ten now.”

It was not a human to issue the tickets for Han’s alleged infractions. It was a network of cameras powered by artificial intelligence, mounted on the windshields of MTA public buses.

And it turns out those A.I. cameras got it wrong. George Han had been parked legally, and he wasn’t the only one mistakenly ticketed.

The problem? According to the MTA, the smart cameras had not been programed to know they shouldn’t issue violations to vehicles parked in the legal alternate side zones that periodically interrupt the M79 lane. The cameras also failed to realize that both the M79 and Bx35 bus routes were still in the “warning” phase of a new enforcement pattern – which means even legitimate infractions should not have resulted in monetary penalties.

“One of the purposes of the warning phase for newly activated routes is to work out any issues before anyone is actually ticketed,” said Tim Minton, the MTA’s Communications Director. “In this situation, there were programming kinks, both in mapping of curb areas and the timing of warnings themselves — all of which have now been resolved.”

In all, the MTA said A.I. cameras on the M79 and Bx35 routes mistakenly ticketed about 3,800 vehicles for blocking bus lanes. More than 870 of the faulty infractions were for cars, trucks and SUVs that were parked in perfectly legal spots.

Johnatan Cuji was one of the hundreds of drivers who received a machine-generated summons – even though photo evidence printed on the summons shows he was parked squarely in a legal alternate side parking zone.

“You can clearly see on the picture that I am literally before the line,” Cuji said. “I was not parked in the bus lane. I always triple or double check every time I park.”

According to the New York City Department of Transportation, the city’s automated cameras – including red light cameras, speed cameras, and bus lane cameras – issue more than 40,000 violations per day. The DOT says a human being is tasked with reviewing every one of those machine-generated infractions, but the agency declined to say how many employees are dedicated to the human review process. The DOT also declined to say how human reviewers missed the hundreds of erroneous bus lane violations issued along the M79 and Bx35 routes.

The MTA said all of the 3,800 mistaken violations have been reversed and any payments made prior to voiding the violations are being refunded. The transit agency also said the software misconfiguration that resulted in errant violations has been corrected.

But Han said thousands of mistakes during the roll-out of the new A.I. bus cameras should prompt more oversight of the vendor who installed and operates the technology.

“Certainly when we’re thinking about the role of artificial intelligence in our society and the roll out of these programs it really has to be done in an intentional way that utilizes some level of common sense before you turn these systems on,” Han said.

A source with knowledge of the matter identified the company behind the onboard bus cameras as Hayden AI, a firm that inked a contract worth $83 million to install and maintain more than a thousand camera systems powered by artificial intelligence.

A rep for Hayden AI declined to answer the I-Team’s questions about the erroneous violations, referring us back to the MTA.

Since the mobile bus camera program launched in 2019, the MTA says A.I. cameras aboard public buses have issued more than 530,000 violations to drivers blocking bus lanes. The transit agency says the initiative — which includes 1,020 buses currently outfitted with A.I. cameras — is working to speed up public buses and change private driver behavior, with only a small minority of violators receiving a second citation.

But the MTA has not provided data broken down annually to show whether the gradual addition of more high tech cameras to the system is resulting in violations increasing year over year. According to procurement documents, the transit agency has agreed to pay a second company $58 million to outfit an additional 1,000 buses with mobile cameras, bringing the program’s total price tag to $141 million.